1) Because of the limited precision of Dx, we calculate it from the given
starting and ending longitudes:
float calcDelta = (lo2 - lo1) / (nx-1); // more accurate - deltaLon may
AFAICT from a picture of the data, this seems to be working.
2) the data is interpolated assuming only that its evenly spaced, apparently
doing a cubic spline fit. that is to say, there is no dependence on Dx. You can
look at ucar.nc2.grib.QuasiRegular if you want the details.
Not sure why you want to write it back out as GRIB1, since its already GRIB1,
unless you just want to convert to a regular grid. In any case, you can set Dx
to 0.14 if you think thats correct. As I said, the actual data interpolation
have not been affected by this value.
On 10/4/2013 1:33 AM, Martijn van der Pauw wrote:
> We currently read in gaussian grid files from ECMWF. When the netcdf
> library has read it in, it has become a regular grid with a resolution
> of 0.1406 degrees. We want this resolution to be exactly 0.14. Is it
> possible to influence the transformation?
> We read in the file using the statement: dataset =
> Kind regards,
> netcdf-java mailing list
> For list information or to unsubscribe, visit: